4927 a 1:1 mixture of 1-hexene and cyclohexane in tetrahydrofuran (with and without sensitizers) was irradiated at the same time. The samples were photolyzed for 168 hr at 27”. The reaction products were analyzed without solvent removal on a 6-ft column of 20% SF-96 on Chromsorb P at 50”. The concentrations of 1-hexene and cyclohexane formed in the two photochemical transformations were determined by comparisons of product peak areas (glpc) with those of known concentrations of the two hydrocarbons. Injection of known concentrations of 1-hexene and cyclohexane indicated 90% reaction after 168 hr. Direct and Sensitized Photolyses of cis- and fruns-3,8-Dimethyl1,2-diaza-l-cyclooctenes1 and 2 (R = CH3). The results of the irradiation studies of mixtures of 1and 2 (R = CHs) under direct and photosensitized conditions are given in Table 11. These experiments were conducted in the same fashion as for cis-1,2-diazacyclooctene (3). In this case, the concentration of 1 and 2 (R = CH3) was 45 mg (0.300 mmol) in 1 ml of solution. After 345 mg (2.08 mmol) of fluorene was added to one tube, the samples were irradiated for 168 hr at 19”. Analyses (glpc) of the hydrocarbon mixtures were performed on a 6-ft column of 20% SF-96 on Chrom-
sorb P at 25” and on a 6 ft column of 5 % QF-1 on Chromsorb W at 45’ without solvent removal. Concentrations of products were determined from comparisons of peak areas with those obtained by injections of known amounts of the four hydrocarbons. Direct and Sensitized Photolyses of trans-3,8-Diphenyl-l,2-diaza1-cyclooctene 2 (R = Cas). The photochemical reactions of 2 (R = Ce&) were conducted as that described above. Into each of two Pyrex tubes containing 2 ml of tetrahydrofuran was added 57 mg (0.232 mmol) of trans-3,8-diphenyl-1,2-diaza-l-cyclooctene 2 (R = CeHs). The samples were irradiated at 0’ for 37 hr after 83 mg (0.460 mmol) of benzophenone had been added to one of the tubes. The photolytic solutions were analyzed on a 6-ft FFAP column at 220’. Comparisons of retention times were made with those of authentic samples.10 Ultraviolet analyses of irradiated samples of 2 (R = C6H5)at different time intervals indicated no cis-trans isomerization of the azo linkage taking place. Acknowledgment. Financial support by the National Science F o u n d a t i o n (Grant No. GP 7600) is gratefully acknowledged.
Polar and Radical Paths in the Decomposition of Diacyl Peroxides’ Cheves Walling,2 Harold P. Waits, Jovan Milovanovic, and Christos G . Pappiaonnou Contribution from the Department of Chemistry, Columbia University, New York, New York 10027. Received January 22, 1970 Abstract: The effect of solvents has been determined on the rates and products of decomposition of several diacyl peroxides, which decompose rapidly by presumed concerted scission of two bonds to yield a mixture of “polar” and “radical” products. Decomposition rates increase markedly in going from cyclohexane to acetonitrile, and the increase is accompanied by smaller decreases and increases respectively in yields of radicals scavenged by galvinoxyl and in polar products (including, in acetonitrile amides and imides from attack on solvent). On the basis of these results and literature data it is proposed that in these systems all products arise via a single rate-determining transition state, and product distributions are determined by the partitioning of a subsequent intimate ion pair-radical pair. The same formulation is suggested for a variety of cases of molecule-induced homolyses, anchimerically assisted homolyses, and radical rearrangements which show large polar effects and frequently strong solvent dependence.
W
hen induced chain processes are eliminated, diacyl peroxides (RC00)2, R = phenyl or primary alkyl) decompose thermally by simple bond homolysis at similar rates which are almost solvent independ e n t . Although there are kinetic complications due to (RC0O)z +2RC00.
+2R. + 2C02
(1)
cage recombination of the primary fragments, a , 4 the chief differences lie in the r a t e of the second step, p scission to yield Re and C02.5 In contrast, peroxides in which R is a secondary or tertiary alkyl or a resonance-stabilized fragment such as benzyl decompose much more presumably (1) Support of this work by a grant from the National Science Foundation is gratefully acknowledged. (2) To whom inquiries should be addressed at the Department of Chemistry, University of Utah, Salt Lake City, Utah 84112. (3) W. Braun, L. Rajbenbach, and F. R. Eirich, J . Phys. Chem., 66, 1591 (1962). (4) J. W. Taylor and J. C. Martin, J . Amer. Chem. Soc., 88, 3650 (1966). (5) For general discussion, cf. C. Walling, “Free Radicals in Solution,” Wiley, New York, N. Y., 1956, Chapter 10. (6) P. D. Bartlett and J. E. Leffler, J. Amer. Chem. Soc., 72, 3030 (1950). (7) L. J. Smid and M. Szwarc, J . Chem. Phys., 29,432 (1958).
by a concerted process involving breakage of two or m o r e bonds. Further, rates are now often highly solvent dependent and increase with solvent polarity.* It was early noted that such peroxides give relatively high yields of ester in which the stereochemistry of the alkyl portion is c o n s e r ~ e d . ~ More ~ ’ ~ recently several have shown that t h e ester is not a primary product, but arises from decomposition of an acyl carbonic anhydride, t h e “carboxyl inversion product’’ (eq 2). Such carboxyl inversion products (RC00)2 +ROCOOCOR +ROCOR
+ CO2
(2)
were first detected in highly asymmetric peroxides13 and a r e generally considered to be t h e consequences of typical electron-deficient rearrangements. (8) R. C. Lamb, J. G. Pacifici, and P. W. Ayers, J . Amer. Chem. Soc., 87, 3928 (1965). (9) M. S . Kharasch, J. Kuderna, and W. Nudenberg, J . Org. Chem., 19, 1283 (1955). (10) H. H. Lau and H. Hart, J. Amer. Chem. SOC.,81,4897 (1959). (11) F. D. Greene, H. P. Stein, C. C. Chu, and F. M. Vane, ibid., 86, 2080 (1964). (12)’-C; Walling, H. N. Moulden, J. H. Waters, and R . C. Neuman, Jr., ibid., 87, 518 (1965). (13) J. E. Leffler, ibid., 72, 67 (1950).
Walling, et al.
/ Decomposition of Diacyl Peroxides
4928
Since these peroxides also yield scavengable radicals most workers have treated the overall decompositions as occurring through competing ionic and radical paths. The goal of the work reported here was to examine the solvent and structural dependence of this competition, initially with isobutyryl peroxide, which had already been studied by Lamb in less detail,8 and then with additional diacyl peroxides. Our general conclusion is that no complete dichotomy can be demonstrated between “polar” and “radical” decompositions, and, indeed, that they both probably proceed through the same rate-determining step to yield a transient intimate ion-paired spin-diradical intermediate.
Results Isobutyl Peroxide. Decomposition rates were determined in several solvents and are summarized in Table I. Individual experiments gave good first-order plots,
Table III. Products of Isobutyryl Peroxide Decomposition, 40” a Cyclohexane
Solvent
Nujol
Scavenged radicalsd 41.5 Carboxyl inversionc 17.9 Ester 5.7 2,3-Dimethylbutane 25.1 Propylene 4.1 Propane 3.7 Other Balance
17.0 24.3 5.1 38.4 9.2 9.8
-
-
98.0
103.8
101.9
Table IV. Isobutyryl Peroxide. Carboxyl Inversion Product and Ester Yields in Acetonitrile
z
carboxyl inversion
-1 4.70, 4.72,b4.5d 4.63, 4.64c 23.8, 24.0,” 24.0d 68.1, 68.5b
a
and the absence of any effect of methyl methacrylate on rate shows that induced decomposition is unimportant in these systems. Our data in cyclohexane and benzene are also in good agreement with those of Lamb.E Several scavengers were investigated in cyclohexane for the determination of radical yields, Table 11. AlTable 11. Comparison of Scavengers in Isobutyryl Peroxide DecomDositiona Scavenger DPPHb Koelsch’s radicab
IO. Galvinoxyl
radicals scavenged 21.4 21.7 32.2 41.2 39.8
24.6 22.3 32.8 43.6 40.4
a In cyclohexane at 40”. Diphenyl picryl hydrazyl. Bis(bipheny1ene)-P-phenylallyl.
a,?-
though all gave satisfactory zero-order plots over small extents of peroxide decomposition, and good reproducibility, agreement between different scavengers is poor. l 4 Galvinoxyl was chosen for further work since it gave the highest values and good material balances. Products of decomposition were determined in the presence of enough galvinoxyl to scavenge all radicals escaping from the solvent cage and are listed in Table 111. Yields of carboxyl inversion product and ester were also compared in acetonitrile at different temperatures and when the peroxide was decomposed by photolysis, Table IV.
0
zester 5.7 2.0 1.7
In presence of benzophenone.
The results listed in Table I and I11 confirm previous observations on isobutyryl and similar peroxides. Decomposition rates and yields of carboxyl inversion and other “polar” products increase with solvent polarity while scavengable radicals and radical-cage products (dimethylbutane, propane, and propylene) decrease. When results in cyclohexane and Nujol (solvents of similar polarity but differing viscosity) are compared we see no effect on rate and little on formation of carboxyl inversion product, but a large drop in scavengeable radicals and increase in radical-cage products: results expected if decarboxylation of isobutyryloxy radicals is fast compared to diffusion from the solvent cage. In acetonitrile, formation of N-isopropylacetamide, isobutyric acid, and excess propylene indicates alternate paths in this solvent for the formation of polar products, and the results can be rationalized in terms of a 0 elimination and a modified Ritter reaction. Additional
C ,
II
(CH3)2Cfi’+li %O
I
+ CH3CN
(14) When compared in the same solvents our scavenger experiments are in only qualitative agreement with Lamb’s.* Thus in CCh with galvinoxyl his group reports a scavenging efficiency of 49.5 % at 45’ ; we obtain 37.4 and 38.9 at 40”.
Journal of the American Chemical Society
29.2 34.3 36.5 7.4
20” 40” 70” 40”, direct photolysis 40°, sensitized photolysis”
Solvent
In
10.2 34.3 10.7 1.1 6.6 0.3 38.7b
-
Conditions
a 0.033 M . In presence of 0.5 M methyl methacrylate. presence of 0.1 M methyl methacrylate. d From ref 8.
24.6 30.2 11.5
Acetonitrile
Yields calculated in per cent based on two fragments/peroxide molecule. Isobutyric acid, 22.1 %; N-isopropylacetamide, 16.6%. Isobutyl isopropyl carbonate. d From galvinoxyl consumption in separate experiments.
Table I. Decomposition of Isobutyryl Peroxide at 40” a Gas Cyclohexane Nujol Benzene Acetonitrile
Benzene
/ 92:16 / August 12, 1970
----f
(CHB)O.CH-N=C /
I1
-OCOCH(CH3)2 (3)
work-up
CH3
+ COO.
(5)
‘OCOCH(CHI)O. I1
(CH3)2CHNHCOCHs + (CHaj2CHCOOH
(6)
4929
examples of this sort of process appear later,15 but here we note that there is good agreement between yields of acid, 22.1%, and amide plus propylene in excess over propane, 22.9%. Finally, the low yield of polar products in photodecompositions parallels previous results and provides further conformation of our suggestion (discussed further below) that photolysis favors the formation of radical products in systems which give both polar and radical products on thermal decomposition. l 6 Other Peroxides. Four other peroxides were investigated in less detail. Mixed rn-chlorobenzoyl acyl peroxides, R = alkyl, have been shown by Denney and Sherman" to give high yields of alcohol, ROH, on work-up presumably uia a carboxyl inversion path. Our results with m-chlorobenzoyl isobutyryl peroxide and also unsubstituted benzoyl isobutyryl peroxide appear in Table V. Although decomposition rates are
since it gave 86.4and 84.8% carboxyl inversion product in cyclohexane and acetonitrile and only a trace (3.8 %) of products hydrolyzing to amine in the latter. a-Chloropropionyl m-chlorobenzoyl peroxide, with an electronegative substituent on its alkyl group should show less tendency to undergo carboxyl inversion. Results here were less clear because of some apparent induced decomposition and because HC1 liberated in the decomposition reacts with galvinoxyl so that scavenging experiments had to be carried out in the presence of pyridine. Results appear in Table VI, and signifiTable VI. Decomposition of a-Chloropropionyl rn-Chlorobenzoyl Peroxide (41") Scavengable radicals
rn-Chlorobenzoyl
Cyclohexane
Acetonitrile
k X 106, sec-1 (41") Scavengable radicals Carboxyl inversion Ester Acidb Other
4.40
103 900 48*
a 12% acid, 40% ester,j later shown to arise from carboxyl inversion product,” combined with other rate data.l0 20% ester, 25% acid.k 30% ester, 32% acid.k d Ester reported,lO later shown to be chiefly carboxyl inversion product.I2 e 61 % acid plus 11 ester on work-up (34 alcohol also present).i f 50% ester,m combined with other rate data.’ Ester, carboxyl inversion product, and diphenylmethane major products, 6.6% scavengable radicals.” 3 7 z ester (with 84% retained stereochemistry), 5 % a-phenylethanol, 673 styrene.O In cc14at 40”, decomposition yields 24 carboxyl inversion product, unpublished work by Dr. A. R. Lepley. R . C. Lamb and J. G. Pacifici, J. Amer. Chem. SOC.,86,914 (1963). k H. Hart and F. R. Chloupek, ibid., 85, 1155 (1962). P. D. Bartlett and F. D. Greene, ibid., 76, 1088 (1954). m M. S. Kharasch, F. Engelman, and W. H. Urry, ibid., 65, 2428 (1953). n T. Suehiro, S. Hibino, and T. Saito, BUN.Chern. SOC.Jup., 41, 1707 (1968). 0 F. D. Greene, J. Amer. Chem. SOC.,77, 4869 (1955).
c
(assumed to arise via processes equivalent to eq 4), and products of attack on acetonitrile. Since most of the literature data were obtained before the importance of carboxyl inversion was recognized and involved product determination after long heating which converts most carboxyl inversion product to ester, we have counted ester and free acid plus any other obvious polar products. l 8 Inspection of Table VI11 permits three generalizations. (1) In spite of enormous variations in decomposition rate (cf. 12 and 14), all peroxides listed show significant yields of polar products, even in nonpolar solvents. (2) In the same solvent peroxides with similar decomposition rates which might be expected to produce radicals of similar stability (cf. 1, 4, 6-10) show widely different yields of polar products. (3) Where investigated (cf. particulary l), total decomposition rates and ratios of polar to radical products both increase with solvent polarity, but the former more rapidly. In short, formation of both types of product are accelerated. Since our own work was completed, a further example of several of these phenomena has been reported by Lamb and Sanderson l9 who have examined the decomposition of a number of ring-substituted benzoyl isobutyryl peroxides, chiefly in cyclohexane. Their results are too extensive to be (18) Esters are also formed, presumably via cage recombination, from peroxides for which there is no evidence for concerted bond scission or the formation of polar type products (e.g., acetyl and propionyl peroxides). However yields are generally low (
